• President Trump met with top AI executives to discuss where to build data centers, highlighting infrastructure as key to U.S. competitiveness.
  • No specific locations or commitments were announced, but the talks signal a push to accelerate domestic AI buildout.
  • The meeting comes amid debates over AI safety and growing electricity demands from data centers.

Trump Talks Data Centers with Tech Titans

President Trump said he discussed with AI-industry leaders where to site data centers during a September 29 White House gathering, framing rapid domestic buildout as central to U.S. technological competitiveness. The luncheon brought together senior figures from Nvidia (NVDA), Meta (META), Google (GOOG), Amazon (AMZN), Microsoft (MSFT), Anthropic, OpenAI, Palantir (PLTR), Tesla (TSLA)/SpaceX (SPCX), and others, according to people familiar with the matter. Reported attendees included Jensen Huang, Mark Zuckerberg, Sundar Pichai, Jeff Bezos, Satya Nadella, Dario Amodei, Greg Brockman, Alex Karp, and Elon Musk. Vice President JD Vance and House Speaker Mike Johnson also participated.

Trump publicly favored AI self-regulation, arguing that existing enforcement bodies such as the Justice Department and FBI already provide some safeguards. At the same time, the meeting occurred amid growing calls for stronger rules after reported incidents involving unauthorized or harmful AI-agent behavior. No specific data-center locations or binding siting commitments were publicly announced, and the White House did not respond to requests for comment.

The discussion sits at the intersection of AI investment, electricity-grid constraints, local opposition, and intensifying U.S.–China competition. Modern AI facilities can draw enormous quantities of electricity; the administration’s July executive order defines a covered AI “Data Center Project” as one requiring more than 100 MW of new electric load. That’s comparable to a major industrial load and can require new generation, substations, transmission, backup power, and long-term utility contracts.

Power and Permitting Challenges

The focus on siting underscores a shift from software alone toward physical bottlenecks: generation, transmission, transformers, networking, land, cooling systems, and specialized chips. Data-center capacity is a practical constraint on how quickly companies can train and serve increasingly compute-intensive models. The administration has already adopted an expansive policy posture toward AI infrastructure. Executive Order 14318, issued July 23, 2025, directs federal agencies to accelerate permitting; identify potential federal, brownfield, Superfund, and—in some circumstances—military sites; use FAST-41 procedures; and consider financial support including loans, loan guarantees, grants, tax incentives, and offtake arrangements for qualifying projects.

Qualifying projects can include those with at least $500 million in capital expenditure, more than 100 MW of incremental electrical load, a national-security rationale, or a designation by key federal agencies. The order also calls for efforts to streamline reviews under environmental and water-related permitting laws, within applicable legal constraints.

Internationally, the strategic backdrop is China. Trump said he did not want to work with Chinese President Xi Jinping on AI governance if it would make U.S. companies less able to outcompete Chinese rivals. He described retaining U.S. leadership over China as the priority. Xi, by contrast, was reported as favoring a more hands-on model of governance and emphasizing human control of AI. That disagreement suggests two competing policy models: the U.S. approach emphasizes speed, domestic capacity, industry flexibility, and strategic competition, while China’s stated framing emphasizes stronger state-led management.

Local Pushback and Market Implications

The main stakeholder conflict is between national-scale goals—AI leadership, industrial investment, and security—and local concerns over electricity bills, water, land use, noise, diesel backup generation, and environmental effects. Trump characterized data centers as positive for communities and argued that limiting them could push investment overseas. But public skepticism is significant: a Reuters/Ipsos survey conducted September 17–20 found 73% of respondents concerned that AI companies had not done enough to prevent serious societal harm, and 55% said slowing AI development would be a good thing.

Data-center siting is increasingly an energy-policy issue, with potential economic effects including construction activity, property-tax revenue, and demand for semiconductors, servers, networking gear, transformers, and electrical switchgear. However, the cost allocation for new generation and grid upgrades is politically contentious if households or existing businesses could face higher rates. Permanent operating headcount is often relatively limited compared with a facility’s land, energy, and water footprint—one reason local communities may question the tradeoff.

The market trend is clear: AI competition is becoming an infrastructure race, not merely a contest over models and software. Expect continued announcements of data-center campuses, utility agreements, and state-level incentive negotiations. Local permitting fights and ratepayer disputes are likely to become more frequent, especially where grid capacity is constrained. Washington will face pressure to define whether voluntary safety commitments are sufficient, particularly after reported AI-agent incidents and delayed model releases tied to safety concerns.

Correction: An earlier version of this article misstated the date of the White House gathering. It was September 29, not September 19.